20 W. M. BALE ON SOME OF THE DISCOID DIATOMS. 



so-called species being differentiated solely by slight variations in 

 the size of the areolation, or by its increasing or decreasing 

 in size towards the margin. All such species, unless other and 

 weightier differences can be found, should be swept aside as 

 spurious. The same remark applies to the presence or absence 

 of a central area, of a central rosette of larger areolae, of 

 bright points at the origin of the shorter radial series, of parts 

 of the surface where the polygonal areolation is replaced by 

 separate circular cellules, and of fine punctate secondary markings. 

 Any of these characters may, of course, be constantly associated 

 with a particular species; but, in many species at any rate, 

 examination of a sufficient series readily shows that they may be 

 indifferently present or not. Indeed, within the limits of the 

 single species C. asteromphalus a range of forms may be found 

 some or other of which exhibit every one of the characters just 

 mentioned, while others show none of them. 



In some respects the size of the valve (i.e. with reference to 

 the average of the species) is a determining factor in the 

 arrangement of the markings. Thus in such forms of C. radiatus 

 as are usually considered typical there are commonly three or 

 four slightly larger cellules in the centre, and the rest are in 

 distinctly radial series. In smaller valves the central cellules 

 are no longer than the rest, and in the smallest forms the radial 

 disposition of the cellules is totally lost. A still more striking 

 instance is found in one of the robust forms of C. asteromphalus, 

 common in some of the North American deposits. The largest 

 valves have a conspicuous central rosette of large cellules, and 

 outside these the areolae are much smaller, gradually increasing 

 in size, however, to the mid-radius. With a diminution in the 

 size of the valve comes a modification in the direction of levelling 

 down the differences in size of the areolation the rosette-cells 

 become smaller, and those next to them larger in proportion. 

 One stage in this series is the C. biangulatus of Schmidt, which 

 is only a normal form of this group, and by no means of specific 

 or even varietal value. In the smallest forms of the series all 

 trace of the rosette is wanting, the areolae are fairly uniform in 

 size throughout, and the centre of the valve is not depressed as 

 in larger specimens, but convex or very slightly flattened, while 

 in many valves the cellules are separate and circular on part of 

 the surface, as in C. perforatus and C. apiculatus. Similarly the 



